Many systems transcribe speech to text. To improve the accuracy of each transcription a system may be trained.